Automatic vessel removal in gliomas from dynamic susceptibility contrast imaging

Area of Science:

• Neuroradiology
• Medical Imaging Analysis
• Oncology

Background:

• Macroscopic vessels within glioma regions can confound Magnetic Resonance (MR)-based dynamic susceptibility contrast (DSC)-enhanced grading.
• Distinguishing these vessels from elevated tumor cerebral blood volume (CBV) is crucial for accurate glioma assessment.

Purpose of the Study:

• To propose and evaluate a novel vessel segmentation technique for DSC-enhanced glioma grading.
• To improve diagnostic accuracy by excluding confounding signals from macroscopic vessels.

Main Methods:

• A vessel segmentation method was developed using clustering of multiple parameters from dynamic contrast-enhanced first-pass curves.
• Glioma regions-of-interest (ROIs) were analyzed from 77 adult patients with histologically-confirmed gliomas imaged at 1.5T.
• Diagnostic accuracy was assessed using histogram analysis, comparing vessel segmentation with and without vessel inclusion.

Main Results:

• The highest diagnostic efficacy was achieved when glioma diagnosis combined vessel segmentation with an initial mean transit time (MTT) mask.
• The proposed vessel segmentation effectively masks pixels affected by intravascular susceptibility effects.
• This approach demonstrated improved diagnostic efficacy for glioma grading compared to methods including vessels.

Conclusions:

• Vessel segmentation based on DSC parameters offers a promising method to enhance the diagnostic accuracy of glioma grading.
• The technique effectively addresses the confounding factor of macroscopic vessels in MR-based glioma assessment.
• This approach provides a reliable mask for excluding intravascular susceptibility effects, leading to more precise tumor evaluation.